1. Field of the Invention
The present invention relates to a focus detection apparatus and an image pickup apparatus that perform focus detection using a phase difference method by obtaining pupil-divided images using an image pickup device having a plurality of photoelectric converters sharing a micro lens, and in particular to a focus detection apparatus and an image pickup apparatus that detect saturation of photoelectric converters.
2. Description of the Related Art
As a focus detection method for an image pickup apparatus, there is known a technique that splits a photoelectric converter (hereafter referred to as “PD”) constituting one pixel of a solid-state imaging device which corresponds to one micro lens into a plurality of PDs. For example, there has been proposed a method in which two PDs (hereafter referred to as “split PDs”) sharing one micro lens obtains respective image signals obtained by pupil-dividing so as to carry out a focus detection process using a phase difference method, and in parallel with this, a sum of outputs from the two split PDs is handled as a one-pixel output to obtain an image signal for imaging (see Japanese Laid-Open Patent Publication (Kokai) No. 2001-83407.)
There has also been proposed a technique that, in order to improve sensitivity characteristics of the split pixel structure described above, reads out charge in one of the split PDs in a non-destructive manner, then reads out a sum of outputs from the two split PDs, and estimates a pixel value of the other one of the split PDs by performing a computation (see Japanese Patent Publication No. 4691930).
In the phase difference computation, a correlation computation result that is not dependent on a subject color can be obtained by performing a computation using a luminance signal. Moreover, using a technique described in Japanese Patent Publication No. 4691930 can improve sensitivity characteristics.
However, according to the technique described in Japanese Patent Publication No. 4691930, because a difference value is calculated by reading sums of outputs from all split PDs after reading out some of split PDs in a non-destructive manner, delay means for providing timing of output signals between split pixels that have been read out is required. On this occasion, providing delay means for respective colors will bring about an increase in circuit size. In one of methods to circumvent this problem, a signal from each PD is converted into a luminance component signal in advance and then input to delay means, and this is thought to save delay means and improve focus detection accuracy.
Incidentally, there is known a characteristic that when saturation occurs in a split PD described in Japanese Laid-Open Patent Publication (Kokai) No. 2001-83407 and Japanese Patent Publication No. 4691930, charge corresponding to the amount of saturation may leak into an adjacent split PD. Because of this characteristic, a signal of a split PD lying under the same micro lens as a split PD in which saturation is occurring causes a phase-contrast image to collapse due to leakage of charge resulting from saturation.
To circumvent this problem, it is desired that saturation is detected with respect to each split PD before a focus detection process, and the focus detection process with consideration given to saturation is carried out. In this case, however, when the above described arrangement in which a signal of each split PD is converted into a luminance component signal in advance is used, a luminance component signal is calculated through computations among other luminance component signals, and it is thus difficult to know an original output signal of each split PD corresponding to the calculated luminance component signal.